Relay choice for scanner

[fcb]

I'm building a scanner to automate the measuring of voltages references. Any thoughts on the suitability of Panasonic TQ2-12V relays?

[Echo88]

For a good scanner you need latching relays, so they dont produce thermal emf voltages during their switching. So yours arent the best choice. I suggest you read the following threads about suitable relays:

https://www.eevblog.com/forum/metrology/low-thermal-emf-scanner-and-ordinary-non-latching-relays/ <- Tests using a TQ2-5V
https://www.eevblog.com/forum/metrology/measurements-on-emf-error-of-switches/ <- general discussion
https://www.eevblog.com/forum/projects/diy-low-thermal-emf-switchscanner-for-comparisons-of-voltage-and-resistor-stand/ <- scanner built from G5A-relays

[Andreas]

Hello,

I would use the latching version of these relays.
I use TQ2-L2-5V see here:

https://www.eevblog.com/forum/metrology/measurements-on-emf-error-of-switches/msg1220467/#msg1220467

with best regards

Andreas

[fcb]

Thanks Echo88 & Andreas,

So it looks like I'm not barking up the wrong tree with the TQ2 series at least.

For my application the relay will only be on for a few seconds every so often so the offset from heating may not be relevant - that said the cost difference (latching vs. non-latching) is not that significant and there seems to be good stock availability.

[martinr33]

The high-end meters all use FETs for switching. For the kind of accuracy you are looking for, you might look the same way. Check out TiN's work on an exotic Keithley scanner card.

https://www.eevblog.com/forum/metrology/volt-nut-test-for-eevblog-readers/

FET switches. Fully isolated - has batteries on the other side of the optos so that nothing can cross over from the main power supply.
Tin's notes indicates that there could be some improvements. Might also be possible to use the packaged switches, although the discrete design might be easier from a battery perspective.

www.mouser.com/ds/2/256/MAX4661-MAX4663-58511.pdf

This isn't too hard to implement. I know that relays are temptingly easy, but they are just trouble if you are looking for high precision. The little optos are very easy to use. If you want to be tricky, you could add series LEDs to the opto drive side. That's very helpful with debugging. Plus, you won't be worrying about coil fields and inductive spikes.

[dacman]

Scanners from Data Proof that are used to scan 732As and 732Bs all use latching relays and scanners from Measurements International that are used to scan resistors all use latching relays.

The Keithley 3700A series of scanners have several different scanner option boards (several hundered channels possible in one meter), most of which have relays (with at least one with FETs). In single channel mode, the offset specification with the FETs is <2.5 uV and the relays are specified at <3 uV, but in dual channel mode the relays are specified at <1uV where the FETs are specified at <2uV.   One major difference is that the FETs are specified at about 64 Ohm where the Relays are specified at about 2 Ohm.

[MisterDiodes]

Yes, the wafer prober test jigs and delicate analog voltage test jigs scanners we use are all latching relay only.  The problem with FET's is you always have leakage onto the common bus - and when you sum up 50 or a hundred or more switches the FETs can be useless depending on what you're doing - especially if the scanner box gets warm and FET leakage goes up.  A latching relay is great isolation from the power supply as well, and an open metallic contact is about as good as it's going to get for a true "open circuit", and typically you'll get less Voffset error in the long run. 

So look at how many channels you're going to use and what effect FET leakage at whatever voltage - that might make a difference on what you need. I'd just use latching relays though to be safe.  Relays are pretty forgiving on voltage levels, polarities, pulse currents etc.

Don't forget that if you're using your switched guarding system typically you'll have 3 switched circuit poles per channel, not just 2.

[martinr33]

For this application, low offset is more important than low leakage as the voltage references should be low impedance. That should make the leakage less important - but I am assuming that this is a fairly low count application.

[Echo88]

The Keithley 7168 uses special copper-lead-JFETs made by Siliconix IIRC (also used as input-JFETs of the 34420A or the input-modulator of E.M. Electronics Nanovoltmeters). So any design using standard JFETs would have to be tested carefully with a suitable Micro-/Nanovoltmeter to ensure that it does produce less thermal emf voltage error than a good COTO-relay (or say G5A/G6A/TX2-relays), which specifies less than 500nV error. I would suggest fcb stays with relays for the moment, a good scannerdesign based on JFETs or PhotoFETs necessitates good thermal design and long measurements.

Another possibility: buy a Keithley 705 with a good scanner-card (<500nV error) from ebay for little money, or any other old scanner (HP, Fluke 7001...).

[fcb]

I'll run with the latching TQ2's for the moment. I don't need to go below 2uV/thermal at this stage.

Drooling over the 'printact' relays in xdevs DataProof teardown, finding myself looking for a manufacturer of flat form spring contacts...

[Echo88]

Indeed those relays are cool, i would buy a few if available. I still believe in OptoFETs and JFETs and will in the coming weeks try to characterise a few of those with a suitable nanovoltmeter. I plan to use a latching solenoid and spring-loaded copper rods (like in the Keithley 147/8 or the Low thermal divider) to built a good zero-switch, so i can measure the amount of thermal emf voltage of the parts i want to characterise.

[macboy]

Some of Keithley's relay scanner cards are specified to < 500 nV thermal EMF per contact pair.  I emphasized pair there because that is important. For minimum offset, always switch both wires through a single dual pole relay. Thermal EMFs will unavoidably be created in the contacts and solder joints, but with everything symmetric and within a single relay, hopefully the temperatures will be similar for both wires in the pair so the thermal EMFs will effectively cancel out.

The main benefit of the FET switching cards is high cycle life and fast switching, allowing dozens or even hundreds of individual measurements per second for days or years on end.